Fire Modeling for Performance-Based Design Project

Two NIST models, CFAST (Consolidated Fire And Smoke Transport) and FDS (Fire Dynamics Simulator), are the most widely used models worldwide for evaluation of the performance of fire protection systems in buildings and are the principal component in performance-based design practice. This project will provide the fire protection community with robust, well-validated numerical models of fire with three components: verification and validation of the NIST models, targeted and novel algorithm development, and continued user support. Verification and validation will focus on including numerous new quantitative comparisons with available experimental data and identify areas of model prediction where data do not currently exist. For FDS, implementation of novel algorithms such as Immersed Boundary Method (IBM) will improve and expand the predictive capability using sub-models that better describe critical physical and chemical processes in fires and can be validated using advanced fire measurement techniques. For CFAST, a primary focus will be on improving the internal structure of the model to modern programming standards to improve robustness of model calculations and facilitate future enhancements to support performance-based design. User support ensures continued impact and provides feedback and guidance for future improvements.

Objective: To develop verification and validation for major existing and new algorithms in FDS, CFAST, and the visualization tool, Smokeview, so that the models can be used to accurately and efficiently predict fire conditions for performance-based design in buildings.

What is the new technical idea? NIST continues to provide the fire protection community^[1] with robust, well-validated numerical models of fire. With increasing use of performance-based design techniques^[2] and evolving standards requirements^[3] for validated fire models, accurate and efficient fire modeling is critical to realization of performance-based design as a viable alternative to prescriptive code solutions.   

With guidance from existing standards, we have now reached the stage in understanding the needs for validation so that we can address model verification and validation (V&V) for both CFAST and FDS in a consistent manner for both models. Continued use of predictive fire models depends on demonstrated accuracy and robustness of the models combined with targeted development based on user needs. Therefore, the present effort consists of three components: V&V, novel algorithm development, and user support. We will focus on including numerous new quantitative comparisons with available experimental data and identify areas of model prediction where data do not currently exist, most notably fully-developed fires in buildings and WUI validation for the most recent version of FDS. Implementation of novel algorithms such as Immersed Boundary Method (IBM) will improve and expand the predictive capability of current fire models using sub-models that better describe critical physical and chemical processes in fires and can be validated using advanced fire measurement techniques. New data generated through solid-phase and gas-phase experiments will guide the development of a framework for sub-models on material burning and soot emission^[4]. User support ensures continued impact and provides feedback and guidance for future improvements.

What is the research plan? The project will be separated into research tasks with the completion of each task being critical to meeting the overall project objective.

Task 1, Model Validation: Much of our validation work to date has focused on the gas phase, with validation for a range of gas-phase quantities for pre-flashover fires. For buildings, additional validation efforts are needed to ensure we have quantified the accuracy of all major algorithms in the models. This would include both natural and forced ventilation, soot deposition, and post-flashover fires as well as extending the range of validation for other variables. For WUI predictions, we will implement all WUI validation cases in the most recent version of FDS^[5]. In out-years, additional validation test data will be collected and evaluated to ensure as broad a base of validation for the models as practical. It is also time to develop a framework for flexible implementation and validation for flame spread as additional research increases our understanding of flame spread on real objects.^[6] In addition, the validation suite for CFAST requires additional cases in order to be consistent with the level of validation performed on FDS.

Task 2, Novel Algorithm Development:

FDS: The development of FDS will proceed along two major fronts – the gas phase and the solid phase. For the gas phase, we plan to make FDS run in parallel on hundreds of individual processors^[7],[8]. In addition to finer grids, an Immersed Boundary Method (IBM) is to be implemented to enable more general geometries to be used within the model. This will allow for direct use of computer aided design (CAD) is the processing of FDS fire scenarios^[9]. The increased resolution in the gas phase will also lead to more accurate prediction of the heat flux to solid surfaces for real materials in buildings^[10].

Smokeview: Smokeview presently visualizes smoke by drawing a series of partially transparent planes^[11].  A better approach that integrates the radiative heat transfer calculation ( involving the term: -k*s) over a longer path length, from the front to the back of the simulation domain rather than from one grid cell plane to the next, eliminates problems visualizing smoke over distances shorter than the grid plane. We will also implement methods for running Smokeview in parallel and in the background to enable the visualization of massive cases^[12]. The end result would be an animation suitable for a presentation.

CFAST:  CFAST is still a vital design tool for the engineering community because its calculations can be run in minutes as opposed to hours or days for FDS, filling a necessary gap in the numerical fire simulation spectrum.. However, CFAST requires a substantial overhaul to bring its source code up to current Fortran standards. While FDS is fully Fortran 95 compliant, CFAST still uses Fortran 77 features that will eventually be phased out.   This initial foundational effort in FY 2012 will result in continued relevance for CFAST and provide the basis for future development. Out-year enhancements will include improved estimation of gas temperature near fire sources (which would improve calculation of heating and ignition of nearby targets), combustion chemistry more consistent with FDS, and the ability to run multiple simulations for sensitivity and scoping analyses.

Major Accomplishments:

Recent Results:   Smokeview is now fully functioning in 64 bit mode, and it can handle cases with many more cells. Major overhaul of key algorithms was implemented into FDS. Improvements were made to the transport algorithm, turbulence model, and combustion/soot sub-models and the results have been documented in the FDS Technical Reference Guide. An agreement was made with United Technologies Research Center to further develop water mist suppression capability in FDS.

Standards and Codes: The project delivers impact through changes to codes, regulations, standards, and practices.  The CFAST and FDS models have transformed engineering practice. Validated and computationally efficient predictive fire models enable both routine and highly complex fire safety engineering calculations supporting codes and standards worldwide.^[13]

• The ICC International Performance Code is completely dependent upon the existence of validated fire models
• The ICC International Building Code recently considered code change proposals whose sole technical justification was the results of FDS simulations (e.g., Boeing Co. simulated large (10 MW) fires in large volume aircraft assembly structures).